def search4Test(keyword, twitterAccount): pages = [] pages3 = gnews.search(keyword) pages.extend(pages3) if twitterAccount: pages3 = twitter.search(keyword, twitterAccount) pages.extend(pages3) pages3 = google.search(keyword) pages.extend(pages3) return pages
def _addTwitterPage(eventPages, word, twitterAccount): if not twitterAccount: return keyword = ' '.join(word['keywords'][:2]) tpages = twitter.search(keyword, twitterAccount) if not tpages: return False tpage = tpages[0] tpage['hash'] = stringutil.calculateHash([tpage['content']]) existed = False for page in eventPages: if page.get('hash') == tpage['hash']: existed = True break if not existed: eventPages.append(tpage) return True return False
def search(keyword, twitterAccount): if not keyword: return [] pages = [] gpages = gnews.search(keyword, large=True) gpages.sort(key=lambda page: page.get('published'), reverse=True) gpages.sort(key=lambda page: bool(page.get('img')), reverse=True) if gpages: pages.append(gpages[0]) else: gpages = google.search(keyword) if gpages: pages.append(gpages[0]) if twitterAccount: tpages = twitter.search(keyword, twitterAccount) if tpages: pages.extend(tpages[:3]) return pages